Fritz/regex
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

bug fixes

Browse source
This commit is contained in:
Fritz Grimpen 2025-02-07 20:58:25 +01:00
parent 0448ebed7a
commit 2d191fbf94
1 changed files with 10 additions and 7 deletions

17
regex.c
View file

@ -134,7 +134,6 @@ bool regex_match(char *haystack, char *needle)
bool bv[states] = { };
// acceptance state
accept = states - 1;
na = false;
// first, decativate all states
for (st = 0; st < states; st++)
@ -154,12 +153,14 @@ bool regex_match(char *haystack, char *needle)
if (needle[st] == '(')
{
// at start of group, look at variants and quantifier
for (var = needle + st + 1; *var && *var != ')'; var = next_var(var))
for (var = needle + st + 1;
*var && *var != ')';
var = next_var(var))
bv[var - needle] = true;
if (*var == ')')
var++;
q = quant(*var);
if (strchr("*?", q))
if (q && strchr("*?", q))
bv[var + 1 - needle] = true;
}
else if (needle[st] == '|')
@ -182,7 +183,7 @@ bool regex_match(char *haystack, char *needle)
fprintf(stderr, "quant %c\n", q);
fprintf(stderr, "suffix \"%s\"\n", needle + (q ? st + 2 : st + 1));
nneedl = start_of_group(needle, needle + st);
if (strchr("+*", q) && !bv[nneedl - needle])
if (q && strchr("+*", q) && !bv[nneedl - needle])
{
// spicy: if the quantifier allows multiple occurences, we have to activate a previous state
fprintf(stderr, "start of group: \"%s\"\n", nneedl);
@ -197,7 +198,7 @@ bool regex_match(char *haystack, char *needle)
// check if current needle is optional
nneedl = next_char(needle + st);
q = *nneedl;
if (strchr("?*", q))
if (q && strchr("?*", q))
bv[nneedl + 1 - needle] = true;
}
}
@ -206,10 +207,12 @@ bool regex_match(char *haystack, char *needle)
for (st = accept; st > 0; st--)
{
// skip beginnings of groups b/c they have been handled before
if (!bv[st-1] || needle[st-1] == '(')
if (!bv[st-1])
continue;
// perform match for active state
bv[st-1] = false;
if (needle[st-1] == '(')
continue;
m = char_match(*haystack, needle + st-1, &nneedl);
q = quant(*nneedl);
fprintf(stderr, "state %lu: \"%s\" ~ \"%s\" => %b\n", st-1, haystack, needle + st - 1, m);
@ -218,7 +221,7 @@ bool regex_match(char *haystack, char *needle)
na = false;
if (q)
nneedl++;
if (strchr("*+", q))
if (q && strchr("*+", q))
bv[st-1] = true;
bv[nneedl - needle] = true;
}